---
title: "Cognitive diagnosis and Q-matrix governance"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Cognitive diagnosis and Q-matrix governance}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(collapse = TRUE, comment = "#>")
library(eyeprocess)
```

The cognitive-diagnosis layer supplies transparent Q-matrix audits, attribute-profile enumeration, and deterministic DINA ideal-response/probability calculations.

```{r}
Q <- rbind(c(1,0), c(0,1), c(1,1), c(1,0))
aud <- eyeprocess_cdm_qmatrix_audit(Q)
aud
profiles <- eyeprocess_cdm_attribute_profiles(2)[, c("A1","A2")]
eta <- eyeprocess_cdm_dina_ideal_response(Q, profiles)
eyeprocess_cdm_dina_probability(eta)
```

These utilities do not replace full cognitive-diagnosis estimation, Q-matrix validation, or model comparison. `fit_eyeprocess_gdina()` delegates exact fitting to GDINA and gates cleanly when unavailable.

Primary package source: <https://cran.r-project.org/package=GDINA>.

<!-- BEGIN EYEPROCESS ARTICLE VISUALS QMATRIX -->

## Visual audit

A compact deterministic Q-matrix makes the structural audit visible. The display concerns declared item-attribute structure and does not by itself establish substantive validity.

```{r m2-visual-qmatrix, fig.width=6.5, fig.height=4.5, fig.align='center', fig.cap='Q-matrix structure for a small deterministic cognitive-diagnosis example.'}
viz_Q <- rbind(
  c(1, 0),
  c(0, 1),
  c(1, 1),
  c(1, 0),
  c(0, 1),
  c(1, 1)
)

rownames(viz_Q) <- paste0('Item ', seq_len(nrow(viz_Q)))
colnames(viz_Q) <- c('Attribute 1', 'Attribute 2')

viz_qmatrix <- eyeprocess::eyeprocess_cdm_qmatrix_audit(viz_Q)

stopifnot(
  inherits(viz_qmatrix, 'eye_cdm_qmatrix_audit')
)

plot(viz_qmatrix)
```
<!-- END EYEPROCESS ARTICLE VISUALS QMATRIX -->